#! /bin/sh
SCANNER="genesys:libusb:002:002"
TXTLANG="eng"
EDITOR="gedit"

echo -n "Project Name: ";
read PROJECT
if [ -z $PROJECT ];
then
  exit
fi
scanimage -d $SCANNER --batch --batch-start 10000 --batch-prompt --resolution 600 --mode lineart --format=tiff

for TIF in out*.tif; do
 echo "Reading $TIF"
 TXT=$TIF.tif.txt
 tesseract $TIF $TXT -l $TXTLANG
done

cat *.txt >> $PROJECT.txt
$EDITOR $PROJECT.txt

PAGES=$(ls -l out*.tif | wc -l)

for TIF in out*.tif; do
 echo "Converting $TIF"
 PDF=$TIF.tif.pdf
 if [ "$TIF" = "out10000.tif" ]; then
  convert $TIF output.pdf
 else
  convert $TIF $PDF
  pdftk A=output.pdf B=$PDF CAT A B output output-1.pdf
  mv output-1.pdf output.pdf
 fi
done

recode UTF8..ISO-8859-15 $PROJECT.txt
a2ps -o $PROJECT.ps $PROJECT.txt
ps2pdf14 $PROJECT.ps $PROJECT-text.pdf
pdftk A=output.pdf B=$PROJECT-text.pdf CAT A B output $PROJECT.pdf
okular $PROJECT.pdf

echo -n "Delete auxiliary files ? (j) ";
read delete
if [ "$delete" = "j" ]; then
 rm $project.ps
 rm out*.tif
 rm out*.pdf
 rm out*.txt
 rm output.pdf
 rm $output-text.pdf
 rm *~
fi